Few studies have explored the associations of reported PA (RPA) with the processes underlying the development of disability. The present study was performed to explore RPA among older persons and its association with onset of functional dependence and mortality. Among a probability sample of 1782 community-living persons, aged 75-83 years, we evaluated the 1021 who reported no disability in basic activities of daily living. Participants were followed for a median of 8.34 years in public registers to determine onset of disability and mortality. RPA predicted mortality in older women (HR=1.77, 95%CI=1.42-2.19) and men (HR=1.65, 95%CI=1.27-2.14) over long time intervals. The effect of RPA persisted among permanently disabled older women, after adjusting for age, baseline vulnerability and grade of disability. Low RPA was independently associated with risk of incident disability (HR=1.56, 95%CI=1.10-2.23) in men. Among older women, the association between RPA and incidence of disability was attenuated in analyses that controlled for baseline mobility function. Thus, the association between physical activity and mortality reflected processes different from those underlying a simple relation between physical activity, disability and mortality. Physical activity was an ubiquitous predictor of longevity, but only for women.